After searching the databases, we identified 2608 references, which were reduced to 2050 after excluding deduplications. Of the 2050 references screened by title and abstract, 73 went forward for inclusion and exclusion by reading the full text. Most studies had a low or moderate risk of bias, except for two studies, which scored below two according to the Jedad score. These were not included in the analysis [12, 13]. Additionally, we excluded one study as it had been flagged with an expression of concern by the publishers [14]. The body of evidence included 43 articles, 27 of which were observational studies [5, 15,16,17,18,19,20,21,22,23,24,25,26,27,28,29,30,31,32,33,34,35,36,37,38,39,40] and 16 of which were interventional trials [41,42,43,44,45,46,47,48,49,50,51,52,53,54,55,56] (Fig. 1).

PRISMA flowchart of the search results and the included studies

Most studies included both males and females in their interventions (between 10 and 60% females), and only two studies were limited to males [52, 55]. The studies were conducted in 16 different countries. A total of 19 studies originated from Europe [5, 16, 17, 23, 27,28,29,30,31, 41,42,43,44, 46, 49, 51,52,53, 55], 17 from Asia and the Middle East [18,19,20,21,22, 24,25,26, 34,35,36,37,38,39,40, 47, 50], and six from North America and Australia [15, 32, 33, 45, 54, 56] (Fig. 2).

Distribution of countries included in the systematic review

Dietary variables were measured by using a variety of instruments. Most studies utilized validated FFQs [5, 16,17,18, 20,21,22,23, 25,26,27,28,29,30,31,32,33,34,35, 37,38,39,40,41, 46,47,48,49,50, 53, 55, 57], whereas the remaining studies used dietary recalls (n = 1) [15], food records (n = 5) [24, 36, 45, 51, 56], a combination of dietary assessment tools (n = 1) [44], a modified Burke diet history interview (n = 1) [54], and not specified (n = 4) [19, 42, 43, 52].

NAFLD was assessed using liver biopsy [30, 44]; imaging methods, including ultrasound [15,